Drill to Detail - Drill to Detail Ep.47 'Business Analytics 2018 Predictive and Best-Practice Christmas & New Year Special' With Special Guest Christian Berg

Episode Date: December 26, 2017

Mark is joined by long-term industry veteran and friend Christian Berg to talk about surviving fifteen years as a contractor in analytics industry, changes he's seen in the market and in how project are approached, the value in getting involved in the community, and in a specially extended Christmas and New Year edition we look back at what was topical in 2017 and what are Christian's predictions for 2018 ... and appoint Christian as Head of our Best Practices Found on the Internet.

My guest in this Christmas and New Year's edition of Drill to Detail is none other than Christian Berg, one of the most experienced Siebel Analytics, S-Base and more recently Oracle BI experts in Europe and a regular speaker at user group events in the UK, the US and Europe. So I've known Christian for about as long as I've worked in Oracle BI experts in Europe and a regular speaker at user group events in the UK, the US and Europe. So I've known Christian for about as long as I've worked in Oracle BI consulting and spent many an enjoyable evening shooting the breeze on topics around consulting, implementing and getting the best out of Oracle and other vendors BI products. So Christian, welcome to the show and why don't you just introduce yourself properly to the listeners and say hello. Well, hello Mark. First of all, thanks for having me,

especially on this Christmas episode. So, yeah, as you said, my name is Christian Berg. I'm doing analytics since quite a while now, especially in the Oracle BI space. I've started doing it in the fabulous year 2000. Still doing it today. Focusing mainly on those products,

but obviously have branched out into other areas more in the open source space and not necessarily other big vendors. I had a small consulting company in Switzerland, which is where I'm based and where I do all my work from. That's where I'm from. So, Christian, looking back at your history and career in this industry, I think you first

surfaced doing S-Base work at a bank who I can't name, but if I described it as a German bank that is run like a British bank, then I think that says the good things and the bad things about it, really. But you worked there for a while with Airspace. And what was your role there, really, on that kind of work there? And what was Airspace being used for on that kind of project? Yes. So in that project, this is basically where we ran into each other first professionally

in terms of working together. Airspace was used for the day trading business, basically for the bookkeeping. And it was obviously like any other S-Base or analytics project at that time, and even, well, 99% nowadays, a project based on S-Bbased cubes that existed already before they started thinking about, let's do some analytics on top of that. And they basically were interested in visualizing what's in the cubes, giving their users the ability to navigate the cubes a bit more easily and more nicely than just through SmartView in Excel. And yeah, as I said, this is basically a trend that has come back many times in different projects where people

start thinking, well, we have these cubes. Why not use them together with our other data? Why not mesh them up? Why not give them to more than just our financial analysts in terms of expenses? So I ran into you there at the bank on that project, and it was back in 2000. And so that was around the sort of time that I suppose Siebel Analytics was coming to the fore with the work I was doing. And I suppose there was talk about bringing it into OBI as well. So what was your background there really with Siebel Analytics? That was your original, I suppose, kind of enterprise bi tool you're working with really wasn't it yes yes so just background background wise um i started out doing uh studies in business administration on process and project management uh kind of realized towards the end that pure management is pretty boring.

So, yeah, I tried to add some spice to it and basically started working with a Swiss-based company on their data warehousing projects as my thesis with them. So I got more into the data warehousing side from there and drifted into the crm side because obviously you can remember end of the 90s beginning of the 2000s crm was the thing everybody talked about everybody was doing and so i ended up in a big siebel implementation on the crm side siebel at that time was just buying Enquire, so I literally arrived in that area when Enquire came up with the fold of the bigger Siebel ecosystem.

Yeah, yeah, the actual first implementation of their Siebel quotes analytics was still using Informatica and Business Objects back at the time. So they VMed two different products. They kept Informatica around, but then obviously replaced the Business Objects part with Enquire. And then over the next two or three years, Siebel realized, hey, this is a pretty neat platform. Let's make it a platform product in our lineup and

that's where this whole idea also started well we have defined data models because at that time they had already you know people software as well why not bring this in we know the base models in the ERP why not make analytical models out of it my star schemas and that's where where they started with siebel analytics applications which is the thing we love uh very dearly still today and uh yeah and then they got bought by oracle and the rest as i say is history yeah i mean i think i think i met you again it was over in Zurich quite a few years ago now. And you were definitely one of the kind of Siebel people.

I suppose the contracting and consulting world in the Siebel world was interesting because it probably wasn't as, I suppose, open an ecosystem as, say, the Oracle world was. But it was very, I would say, lucrative. I don't know what way of putting it. But certainly it was a nice niche to be in and then uh or is that not the case well it's it well you mentioned two things there uh one is openness and second is lucrative and um just just to talk a bit about about the first one because i think a lot of people have a huge misconception about this in their head as well. So Siebel was really the opposite of Oracle in terms of communication, community, etc. I mean,

it was closed, closed, closed, closed. Even just to get basic documentation and whatnot was a pain, really. So they were completely secretive about everything, didn't share. So as a result, there was hardly any community. People were trying to build them, but it was hard because nobody really got any new information. You're presented with a new product and basically that's it so for that um oracle actually really is a lot better i mean it already was when they bought siebel but nowadays with all their different activities in the community space uh and their open source activities definitely definitely definitely a leap ahead in terms of lucrative it never really, you know, where can I cash in the most. For me, right when I started being exposed to Enquire, their metadata engine and their common enterprise information model with the RPD was something that worked for me.

It just made sense. Looking at the other tools at the time, yes, Business Objects has their universes and whatnot, or had at that time, but it just wasn't the same thing. So this common transformation layer and really getting the information together, I think that was so far ahead of its time that it was just completely misunderstood back then and that's the reason why some people think nowadays i mean we're in 2017 for crying out loud that they think well this is old technology it's around since almost 20 years so it must be

you know outdated and not fit to our requirements. But I think, well, if you look at it, it actually does a lot more than you even require. So the power of the thing and the flexibility of the thing was just always something that spoke to me. Yeah, I mean, I suppose in a way um the way technology has moved on um and and what we're using today in some respects is is is kind of more powerful but in other respects there were quite a few kind of old things that have been almost forgotten really and and uh you know the bi server is one of those things there um and we'll get on to that and that's quite

interesting sort of topic area really as well and so christian you're you're now working um mainly with obie which is obviously how most people know you now um and so christian you're you're now working um mainly with obie which is obviously how most people know you now um and um you know you're you're so you have your own um i suppose you're a contractor aren't you so you're a consultant but you work for yourself in this kind of market and i suppose one of the first things i want to explore with you really is is i've described you in the past as one of the kind of survivors you know you've been doing this you've been working as a contractor in the in the analytics market for about 15 years or so now um you've you've been you've survived things like you know the transition from uh from uh seaboard to oracle you've survived um

the in quotes interesting kind of integration of s space into obie you know you've survived things like the indian outsourcing um sort of phenomena you know you're still here as well um so i think that's really i want to go through you know your perspective on that a little bit really but first question i'd ask you really and it is why have you why have you always been a contractor as opposed to you know joining a big consultancy where you have more impact or even formed one yourself okay well you just said something very interesting. The question in itself is interesting per se, but you said, why haven't you joined another ability to have more impact when you're part of a big organization and and that's why this isn't uh these are the other reasons why i never joined up so a i never found a company so far that uh that was interesting to me in terms of culture and ethics so some are pretty cool some are pretty you know hip and trendy but well it has to it just has to work and the other thing is as time progresses many companies become just another consultancy got the normal road of well we have to look at the baseline of the company we have to make profits so what

we'll do is we're basically gonna reduce our quality of resources get more young ones get more cheap ones just place them at the client and then they can learn at the client and that's cool and which is just something that doesn't work for me so if you arrive at a client and you don't know what you're talking about and you more or less have to spend two months at the client just learning the product well you can do it with any of the big fives you will get the same from everybody well sure that doesn't interest me in the slightest i'm actually interested in the slightest. I'm actually interested in getting things done for the client, in getting benefit for the client,

getting the right solution. And so for me this also means, which is in terms of management obviously something that you shouldn't do, but sometimes just walking up to the client and saying, listen, you're completely on the wrong track here. You shouldn't do this, even in less flattering terms, and say, well, if you decide to do

it like this, this is your problem, but I won't put my signature under this. I suppose you've always got much more control over the quality of the end product, haven't you? And you've always got a much more direct relationship with the customer um which is which is good i remember speaking to yeah i remember speaking to adrian ward on here a year ago or something yes and and so my point here was made at the time was that the other thing about being a contractor going in somewhere is that you're in there for longer than five minutes you're in there you know the advice you give has to kind of like i suppose in a way work over the long term

and you're there to actually see the consequences of it and you're and you're there to actually adapt your advice and and you know your advice after that based on actually the reality of the situation there sometimes yeah sometimes sometimes this is the case but um i don't do exclusively long-term contracts. I do short-term assignments as well. So it can be the case, but I think it's more about, yeah, as you said, getting the right result and not saying, listen, here's a list of requirements. Sit down, do it.

Cheers, thanks, bye, and then you're gone. So this whole resource switching just won't work. Yeah. Especially in the analytics space. Especially in the analytics space. It's not like you're writing an application, button A does this, button B does this, and you can write a script that does all, every time it does the same thing, you know, expected result, followed process, done. So do you think, do you think that

analytics projects are a special case then when it comes to development? Is that why perhaps they have survived the great movement towards outsourcing over the last few years. Oh, yes, definitely. Yeah? Tell me how, then. Oh, yeah. I mean, the normal, well, let's say, outsourcing project just does what I said before. You know, you have some people sitting there at the client being the only, quote, local, unquote, resources in the whole project. They're doing nothing but talking to people, writing work documents, producing paper, basically. And then they'll say, all right, thanks so much.

We'll hand this over to our outsourcing development team now. They hand it to a huge team somewhere. They have no idea of the client. They have no idea of the business. They have no idea of the requirements. They have no idea of the business. They have no idea of the requirements. They just get some paper. Most of the time, they don't even have any idea about the technology. They'll just say, oh, yeah, yeah, this week I'm doing X.

Next week I'm doing Y. And then they start producing stuff. And then stuff comes back to the client. And the client says, well, this is not what we wanted. Integrator says, well, you should have specified it more precisely. It goes back, et cetera, et cetera. So this whole cycle of correcting, misinterpretation, misunderstanding, and it's already hard to do in a normal application world

where you can very clearly define things, as in saying, as I said before, this button does X, this button does Y, and then in the end, I have to have, I don't know, a new purchase order. So in the analytics space, everything changes all the time. Data flowing in

changes your reality, your models can change, new things can come in, all things fall away, et cetera, et cetera. So the added problem there is you actually have to understand what are they trying to achieve, and trying to achieve not as in this has to be blue, this has to be yellow,

but what's the business goal that needs to be achieved. And this is very hard. I mean, to put it in perspective, I was at a client once where they decided to go with an outsourcing partner. Literally, they had set aside an additional budget of 50%, the initial budget, just for bug fixing because they said, well, we know this from other projects that we've done uh with this kind of partners and this is just how it works period

i mean i was flabbergasted this is to me this is incredible project management this is just this is just rubbish and the other thing yes yeah the other thing the other thing that I wanted to mention is, if you talk about analytics, you talk about not just knowing, for example, Java and being able to program Java, or knowing how to do Oracle Analytics, or knowing how to do Excel in the worst case. It's about actually comprehending what you do with the data and comprehending what works

with the tools, how and choosing the best solution. I mean, as you know yourself quite well, you can't just say, I have this tool. I solve every single requirement I have with that tool. And especially when you look at the Oracle, let's say, ecosystem, this is true like times 10 to the power of 20. Because you got, imagine you just buy the normal Oracle Analytics product suite. Minimum, you have Oracle BI, BI Publisher, you've got S-Base, and hence you also got Smart View. You've got data visualization. So there's already five different tools to do something with your data.

So which tool do you use for what? And this is the thing about comprehension, if you look at the, for example, Oracle Developer Community Forums, you'll see tons of questions that are, I need to do, or I have requirement X, I need to fulfill it with Oracle BI. How do I do it? The first,

let's say, five answers are, you're using the wrong tool, and they're just being obnoxious and saying, no no this is the requirement the client told me to do this and that how do i make it work this is just completely wrong you can't you can't yes yeah you can't you can't uh you can't you can't just go ahead like this yeah you always have to say well again, again, what's the requirement? What are they trying to achieve?

And the client can tell you how to achieve this if he's not a master of the technology, because otherwise, why did he hire you? What's the point? Otherwise, he could do it himself. It's up to you as the contractor, as the actual developer, to say, okay, listen, if this is what you're trying to achieve, the more optimal road is to use another tool or is to go down another path. As we said, I mean, do you remember that famous slide

that you made of me back in Israel? One of many, probably, yes. About exporting to Excel. Exactly, the one about exporting to Excel. So as a context for the listeners, Mark and me were at a conference in Israel and I was talking about basically feeding the Excel monster from Oracle Analytics.

Mark made a nice slide where I was presenting and behind me it just said, export to Excel and I will kill you. The background to that was people are using the whole analytics platform as a basically pass-through to just push millions and millions of rows through CSV or Excel files to the desktop of people, where I'm saying, yeah, it can be a requirement to have this information, but why do you have to go through the whole metadata layer to the presentation server to a browser? So, I mean, network traffic and whatnot, humongous.

Instead of saying, well, yes, it's a requirement. We can do this at a very low level and just feed the Excel monster directly from the data sources, whatever that data source is. Every data source on the world can do this nowadays, just dump out data. So again, taking a step back from requirements and thinking about what are we trying to achieve and how do we do this? And this is something that for me sometimes is, well, I'm taking more or less work away from myself because I'm saying, you don't have to do this in an analytical platform. You can do it with other tools far more easily, far more efficiently. And this comes back to my initial point about why didn't

I join certain companies? Because, you know, we always work for somebody and that somebody at some point will say, listen, gotta stop doing this you gotta just start doing what the client tells you to do literally with the tool that we're there for basically to to maintain closing the loop yes yes i mean so so i mean there's a there's a there's a lot of unpacking in those statements there but i mean it's so you i suppose the quality of this really is is you know you can be very true to yourself if you're a contractor you can be very and that's not a derogatory sort of way of putting it you you can't you you have you have your kind of core mission your core ethics and you have your core thing you believe in and as a contractor you

can deliver on that day in day out but i guess the other side of it is it's a tough business to be in if you're also having to be selling at the time selling yourself and so on and and you know I suppose as trends in the market recently have happened with things like cloud things like outsourcing you know it was what's the contractor business like these days and in a way is there still a market for technical consulting or you know first of all what's the contractor business like now is it still as I wouldn't say lucrative but is it possible to exist in that as well or what now uh it's changing definitely it's changing it's getting it's getting harder um and for well for for several reasons you know with the whole change to to cloud and uh some clients really some getting more interested in also using open source tools and whatnot you have you have these two

let's say lanes so one is uh let's start with open source oh if you if you say your clients are getting more interested in using open source tools and obviously you have a huge opening for contractors. On the other hand, cloud in the first step I would say is more something where people or let's say end clients think yeah this will get rid of our consultants, this will get rid of these contractors. Yes and no of the classical ones, of the old school ones, of the ones doing pure development, like hacking code, moving data from A to B, using, I don't know, a data integrator or talent on a day-to-day basis.

But it does open the door for a completely new approach, for this whole approach of saying, well, finally, and just our logo everywhere, the whole GUI customized. We want new buttons that do stuff. So customization basically has nothing to do with the purpose of the tool. Basically buying a car and saying, actually want i want vases as wheels i want a palm tree to grow out of the roof and i want the engine to to be replaced by a clown clown car you're like yeah yeah okay this is nice you're declined you're paying for this i know you insist on this very very heavily and this is actually a cool thing about a lot of the cloud

tools out there they just don't allow you to do that they're saying this is the functionality this is what you get this is how you use it period yeah so so have you found though that with this i suppose changes in the market where people are more used to self-service tools now and there is perhaps a greater level of awareness of analytics and I suppose kind of data discovery and tools like Tableau and so on. Have you found that actually customers have been able to do this themselves now and there is less of a need for an experienced analytics contractor? Again, it heavily depends on the customer. What I've seen, let's say, these last 24 months

is customers who have, again, almost everything. They have Oracle Analytics somewhere. They have another department using the Hyperion EPM products. They have another department using Tableau. They have another one using SaaS. They have another one start doing things with R. And it's a very heterogeneous environment

and people start less and less to talk to each other and building their own solutions. No, ours is better. Ours is better. So again, it comes down to all these tools have their reason for being. But what about general awareness?

So maybe it's kind of like a misconception on my part, but people seem to be more able themselves these days to do analytics. What do you think about that? Yeah, it's not a misconception. There are people becoming more and more apt at doing things, definitely. The problem is, and this is where I rejoin the point that I wanted to make before about corporate culture and change management. The problem is structures as a whole need to change. There is a huge, huge, huge advantage nowadays if you're in a company that is very data driven and very

analytics driven and in in many cases also unfortunately it has to be said younger so younger companies by default are a lot more open to this if you're in huge established corporations and i work mainly in the finance industry so as you can imagine these are very big very established uh structures so it can be hard there can be there can be these um these nests of real let say, data analytics and affinity in the corporation, which are really good, which are really driving things forward, but which are hampered by the general and more global structure and environment of the corporation. So, yes, there's definitely, let's say, an awakening of things happening in companies. But again, hampered by silo organizations,

hampered by, and this is the other thing, everybody talks about data lakes and whatnot, and free access to it, free the data, yeah. Reality, unfortunately, is quite different. Because normally the people hoarding the data and the people actually doing things with the data are completely separate and the ones are the clients of the others but the first don't really have to the resources or the time to so basically analytics and all these newfangled things where you say well yeah data let's do it, are being hampered by actual access

to the... And culture as well. I mean, I was, we were both in the keynote at the recent Oracle User Group Conference in Birmingham, and it was a guy from Oracle was talking about some really exciting, interesting new technologies and products that were coming along

from Oracle around things like using machine learning to help in line of business applications and the talk was about becoming a data-driven organization and I did a talk directly afterwards to talk about what I've been doing recently and and you know it was ostensibly the same topic matter but what struck me at the end of it was how you know you can put all the new technology you want to into a company but if the culture isn't there to actually use numbers and use analytics and data to drive that business layering all this technology on is not going to make a difference really and it's all about culture it's all about culture it's not necessarily about age or whatever but but i mean obviously i think

there is a there is a there is a correlation there but it's about culture it is it is it is about culture and then and as i said it's just that the existing established corporations normally have a bigger let's say hurdle to cross there so but yes the culture the cultural impact and and that's what i meant with um also clients that i walk into nowadays have not just one or two tools, they have 15 tools. Because every department has what fits their boss's culture and their little cultural stuff. So it is an issue when your corporation doesn't have an overlying exposure and strategy for data and analytics.

So if the company as a whole isn't geared towards this, it's very hard to change things on a more than small scale. So something that was interesting in the conversation I had with Carsten Banger last week was I asked him, he was the guy who was behind the bi survey and i asked him a question which was are you finding so most of these most of these kind of bi um you know market analyses you see coming out they look at actually what is being sold net new and they and they they basically derive trends from that but i asked him uh how the organizations that have bought these enterprise bi tools like cognos or oracle or

whatever and there's been a lot of time taken to build out these semantic models and semantic layers they were meant to address this kind of issue where um there were lots of different kind of silos of data and there were lots of there were lots of i suppose little individual initiatives and numbers never never added up but what was interesting when i spoke to him was he said that still most new projects involve a new bi tool, not necessarily reuse of what they've got. Is that what you see, or do you see these investments being paid off now in semantic layers and enterprise BI tools? Sometimes. Sometimes, yeah. It really depends on the comprehension of the topic inside of the corporation. I've seen it in many places, unfortunately, go horribly wrong, where corporations move from tool A to tool B, let's say from a more reporting tool to a more analytics tool.

And it just reproduced what they had in the new technology without using any of the semantic features, without really using the metadata layer, et cetera, et cetera. So basically just, oh, yeah, we're doing new things. And the fact that we have new technology will solve all our issues and that's and that's just the thing no it doesn't you can move you can move from i don't know sap to to oracle you can move from microsoft to tableau you can move from from from from excel to one of the let's say open source source R distributions out there. It doesn't solve your issue.

You have to actually change the way you work with data and change the way you think about data, change the way you handle it. So, again, a very bad example that I ran into recently is a company using SAS and just, well, SAS as their more or less self-service data discovery tool with the problem, the problem being the actual concepts behind not being understood and literally people transferring terabytes and terabytes of

data using dozens of cores and hundreds of gigabytes of RAM to do the most basic things instead of first thinking about what do we try to do here let's trim down the data to the bare minimum then we have our output then we correlate to it with the rest of our data. So, you know, procedural thinking about data as well. And then, obviously, well, overhead and license costs just skyrocket. Because you're just using it wrong. For things that you could do with a pure, let's say, Oracle database with R integrate.

Just run it on the database, push the output to whatever front-end tool you want to have, and then do stuff there. Well, in terms of licensing and the money involved, it's a lot better usage. But just coming back to your other question, because you mentioned all the big data ecosystems, and a friend of mine reminded me yesterday, or the day before yesterday, about a quote from a guy

called Dan Riley and he basically said, big data is like teenage sex. Everybody talks about it, nobody really knows how to do it. Everybody thinks everyone else is doing it. So everyone claims they're doing it. So it's been a while back that he said this.

It's actually four years ago. But I remember this quote, and it's actually a very, very pertinent quote still today. Because yes, you have these companies that are really embracing big data, that are embracing data analytics. But then you have the big chunk that is the rest of the market, which isn't necessarily just 20%, but which is still about 70, 80% of the market. Everybody says they're doing it. But if you go into the corporation and look at what they're doing and how they're doing it, then you say, well, A, you could have done this with the technologies you had before, A. B, you more or less implemented new technologies just to do this

for nothing as a fact. And C, yeah, this isn't really what the whole big data and data analytics idea is about yeah i mean i think there's a couple of bits in there really i mean you've got i mean there are companies out there doing big data and and using it to benefit customers and to benefit obviously the company itself company itself and maybe society and so on. I mean, so working in the area that I'm working in now, it's very clear that there are companies doing that. You've also got companies out there that are playing around with technology. And you could phrase that as being, I suppose, doing research R&D and trying out new things

and perhaps using technology in a way that isn't really needed that way and so on there and you've got companies there that have no idea but I definitely do think there is a difference in you look at these data-driven kind of high growth SaaS companies you look at companies that are I suppose in a way personalizing on a one-to-one basis everything they do for their customers based on knowledge of all their actions and so on. And they are doing well. But I guess not every company is a big data company.

And not every company is a startup. And so I think it was ironic, I think, with my move out of Oracle into what I'm doing now is the technology that actually I think a lot of us are still kind of playing around with in the Oracle world has been left behind years ago. And it's all now kind of cloud-based distributed query and storage platforms. It's not about HBase. It's not about kind of Hive and that sort of thing. And I think that certainly it surprised me going back to the Oracle event the other week where people were still talking about connecting this to Hive and using this with Spark and so on yeah and I think I mean don't it's very easy to get caught up in the technology um but it's I think the ideas

are good that are out there but as you say you know not everything is a big data problem and not everyone's doing it and and so on really um and what about getting looking at yeah I mean looking at looking at you know you're used to kind of working with things like semantic models and and you know bi servers and so on and yet these days people are using tableau and they're using um you know click or they're using uh very kind of simple open source tools do you think we've kind of gone back is it a bit like the fall of the roman empire you know when if you think about the the world of enterprise bi was the kind of roman empire everything was incredibly well put together and organized. And then you have

the dark ages of the Germans running across the German plain, smashing everything up and everyone living in mud huts then. Do you think we've reached that point now? Or what's your view on that? Unfortunately, let's say Enterprise

BI and data models, you know, generic, not generic, but corporate data modeling. Governed ones, yeah. Yeah, governed ones has never reached the stage of the Roman Empire. It has never gone there. Let's say as an equivalent with the Romans, they would have reached maybe Hegemony about 150 BC, and then the Germans came. So they never took, the whole idea never really took hold in, well, never widespread. It was never that widespread.

So everybody talked about it. But again, very few people did it. And so they thought that with the coming along of big data and data analytics and all these automatic data discovery tools and whatnot, it would be a Hail Mary. And then you have the other tools like Tableau, which were more initially departmental tools. So going in certain departments, solving their issues, doing their little thing, and then growing from there to the outside. So more or less like a cancer, you know. It passes from place to place. No value judgment there though.

No, no, no. I mean, again, every tool for its purpose. You don't have to use an oracle analytics to kill every single tool because sometimes it's just too humongous. So no, my point is, and I mean, if you want to stay in the Oracle space, even Oracle with data visualization

and data visualization desktop has these tools now that solve these little issues at certain things. And, you know, everybody does more or less the same in terms of value proposition. But I guess people saw Everybody does more or less the same in terms of value proposition. But I guess people saw or realized, well, there's work involved. We actually have to think about these things, and it takes time to think about these things. And then, you know, if you add to this the normal inertia of a big corporation with different departments, you know, silo organization, You have to wait for the data to be there.

Then you have to wait for the data to be loaded. And then, oh, it's a DBAs, but they're actually used up in a different project. And so we have to wait another month. And then now we have a slot. And in three months, we have a release window where you can get your results. Thanks, cheers, bye. And people were turning towards more quick shot tools like Tableau to say, we need something fast.

We need to do something. And, you know, corporate information modeling, be damned. Let's just do it. Well, yes, of course, it does work. But I'm just mentally jumping ahead a little bit in the discussion and also referencing something that Carson said in his episode. This is all fun and games, but especially here in Europe, corporations will have to work with another reality that's coming up, which is GDPR. centralized reporting tools, analytics tools, little data silos, data ponds residing in glorified Excel sheets, they will have a huge problem legally next year.

Well, actually already now, but next year you can be fined. So I think nobody actually thought about something like this happening. These going back to the desktop, going back to silos of information just because you're quicker trend will definitely have a legal issue. And I think there the whole cloud aspect can help corporations unify these two requirements. On the one hand side, speed to market or speed to your own internal clients. And on the other hand, traceability, always know where your data is, always know how it is being used,

who does what with it, when, who had access to the information, etc. Because it kind of merges these two worlds. So the, let's say, corporate data modeling world with things like the bi server they had the whole tracing capability they they they had the ability to track back who does what when with data how does it get transformed according to which rules etc etc and then you have the more uh wild west of desktop tools where you don't have this but you have a huge speed to market and so i guess these two will i mean they're already

merging luckily already merging again and we're coming back to well all these concepts but more again centralized but not centralized at the client but centralized in tools yeah yeah i mean let's go on to gdpr a bit later on because i think it's something to talk to you about and get your take on that um but another another analogy so you know back in the 60s back in the 70s we had the sort of people landing on the moon and we had uh concord and and uh and and a lot of kind of technology that make you that would make you think that we would be, you know, all living on Mars now and with jetpacks. But actually we're not, and we're stuck in traffic jams and we're playing kind of Candy Crush Saga on our phones.

And if you look at, if you take the analogy with that in BI, you know, 10, 20 years ago we had OLAP tools. And, you know, we would have a sales analysis application where you got your responses back in sub-second. And the data was was was was well organized into hierarchies and so on and actually now we've got kind of like data visualization mashup tools and and nothing makes any sense at all do you think you know do you think that we have regressed a little bit when it comes to kind of i suppose uh the way we do analytics on the

desktop or or is it all kind of good now in In certain ways, in terms of structure and procedures and whatnot, we definitely have gone back. But I think this is, I think we're really still, even though this has been going on since years, I think we're really still in a very much transitional phase. You mentioned the dark ages before, you know. And then all of a sudden came transitional phase. You mentioned the Dark Ages before, you know. And then all of a sudden came the Renaissance.

And I think we're really kind of like in the Renaissance right now. People are starting to wake up. Obviously, in the Renaissance, they tried out many things. Many things that went horribly wrong. I mean, Da Vinci was a genius, but a lot of his stuff was just, yeah, that ain't work.

And I think really in a Renaissance in terms of analytics and data handling, we had a very, very structured, let's say, Roman Empire approach, the barbarians came, desktop tools came, and yeah, data modeling is dead, and the semantic model is dead, and who needs data structure? Everything is just the same. Yes, well, that didn't really work out. So I think we still need some time. We still haven't reached, let's say, critical mass for everybody to follow. We got a good chunk, yes, of people starting

to think differently and especially, you know, new generations coming into corporations. They think, let's put it like this, they think differently out of the box. We don't have to change their mindset. We can actually work with them right away. We don't have to say, okay, yes, this is how you did it before, but look at these advantages, look at these changes, which we have to do with a lot of established departments and people working in key positions.

So this is the good thing. Change takes time. Really, it takes time. It's good that there are people rushing ahead, and we need these spearheads. We need the tip-of-the-spear type of behavior that's penetrating new areas, that's going deep into things, where they say, well, we we tried it it was a great idea

it didn't work so you don't have to take this experience again it's like you said a while ago you know six seven years ago people were were trying to oh yeah we have to build our own hadoop cluster we have to build our own hadoop cluster and who builds their own hadoop cluster nowadays yeah yeah I do I used to have one in my garage a little while ago so uh but this um so so what about another another another kind of big story arc that's been going on now in our world over the last kind of 10 years is the unification of OLAP and relational so I remember but I had Mike on the show the other week and Mike was was with a product called Discover for OLAP,

I think it was called years ago, when Oracle tried first to integrate OLAP, as it was technology, multidimensional OLAP, into relational tools. You've had the whole, dare I say, debacle of Oracle doing that with S-Space over the years. In fact, you borrowed a slide off me at the last conference where I had a kind of like a potted history

of every time Oracle said, this version, it will all work. Do you think, I mean, maybe not necessarily kind of pinning it on Oracle, but other vendors as well. Do you think it's a folly to try and unify OLAP relational or do you think there's a point to it?

Or, and if there is a point to it, do you think it'll ever work? You know, as someone who's been very much at the front line of that, what's your take on that the front line of that, what's your take on that? So, I mean, to put it blunt, yes, there is a point, and yes, it does work.

The question is just how far do you want to push it and how far does it make sense to push it? I think this is the thing. How far do you want to take it? Because we all know multidimensional sources, just like relational sources, just like key value stores, just like any other storage technology out there, is there for a reason because it fulfills a certain purpose. And for multidimensional sources, this is clearly measure hierarchies, nonlinear aggregations, stored members, etc. So these are the key value propositions of a multidimensional storage compared to a relational one. Because in a relational one, I mean, yes, you can do all those things,

but the effort is just too high. Also, you can just chuck all the data into, let's say, an unstructured storage, just have it lie somewhere and say, yeah, we'll just build lot of processing power, but still in terms of response time, since these things are made to use this just as it is, and are made to store it just as it is, you will never get there. You will have a huge amount of effort. Imagine you write something to a cube once during load and then everything aggregates as it's defined. And on the other hand you have dynamic access through an analytical tool and the analytical tool has to do all the logic well it does it every single time the thing runs so it's very suboptimal so these these these cubes since they're

mainly and I have to stress this point, really, they're mainly in their usage, they're more like applications rather than analytical platforms. So they're used to, for example, do planning and budgeting. They're used to do scenario calculations. You're not there to have 500 different users, access them constantly with changing dimensionalities, drilling down to the nitty-gritty detail all the time and then dumping it to Excel or whatnot. I mean, in terms of data pulling from a cube,

this is just, yeah, pointless. So the integration works. It does make sense. The question is just what do you want to achieve with it? If I build management cockpits on top of it that summarize the information to our management, then on the other hand,

we have our hardcore users that are going into these cubes, into the real OLAP world with proper OLAP tools. Yeah, it doesn't make sense. And honestly, so far, if we come back to the more modern world

with the new tools, the whole big data ecosystem out there, there aren't that many tools that are geared towards this simply because the technologies underlying it, like multidimensional cubes, are not really something that is being represented in the big data world because it's old technology. They're like, yeah, yeah, you know, cubes. Who wants the cubes

to do cubes in the cloud? Except for the people that have been doing it all the time anyways, like Oracle with their S-based cloud offering. So it's the same thing. But again, you have these storage technologies. I mean, how often have you heard in the last, let's say, 20 years, how often have you heard the phrase, the database is dead? Yeah, I mean, I think it's, I mean, nobody, I suppose, that's the phrase that begs itself to sort of say, well, actually, it's not. But nobody leaves college today to be a DBA. And, you know, looking at the kind of startup world that I'm involved in, there isn't really a role of a DBA.

You have infrastructure engineers, you have data that's stored in the cloud and therefore doesn't really need to be, I suppose, backed up for DR purposes. And databases are... I suppose the background to that statement is that databases are actually more of a kind of bit bucket. You just chuck stuff in there

and the logic goes into the application that is happening now and and i suppose the care and maintenance of a database is is kind of less of a less of an issue but people still need to store data somewhere and have it in a way that's structured and and and safe and so on there really but i mean it's certainly less i think the data the dba in an older organization would have a massive influence and power. And I think DBAs now, certainly they would manage thousands of databases, not just one. And the database is more like an infrastructure component as opposed to a first class, this is a thing that we kind of worship at the temple of really. And I think in a way what's interesting is things like the Oracle user group.

There is still a massive community of DBAs there. Oh, definitely. Definitely. For most of the conferences that I go to, it's still the biggest part, which is amazing, actually. Is it not a load of old men

talking about technology they used to work with, though? Is it not something... I don't know. It's an interesting topic, and going back a little bit to to i suppose uh grand unification of olap and and so on there i mean that that was an interesting area and i think that one of the challenges maybe again for you as a contractor actually as a contractor it's probably easier for you in that you could go a client could buy let's say for example tm1 and cognos and expect it all to work together. And as a contractor, you can say, it's not going to work.

Whereas I think as a partner and as a kind of consultancy, that was always an interesting kind of area to be in because you didn't want to embarrass the customer in front of his bosses. But also, and there was obviously money to be made out of companies that bought a suboptimal thing and then you get to make it work but i guess that's a nice thing well i suppose my argument was always you know my argument was always sometimes a thing is bought by somebody above somebody else and they this is the technology they have to work with and so they want to get somebody in who can try and make it work for them but you know that's a that's that's a kind of soulless job in a way sometimes because you're batting with kind of one arm behind your back,

whereas as a contractor, you can be a lot more blunt about that really and walk away from it potentially as well. Most definitely, most definitely. I mean, you're basically describing the state of the market in many corporations. It's being bought by someone who just takes a decision based on some very nice sales slides and whatnot and the lucrative contracts you can make in the background. But then people actually working with it are stuck with it, as you said. Yes. Yes. Okay. Let's move on because I'm conscious of time. So let's talk about community.

So I remember several years ago, I was running a conference and you couldn't make it because your your girlfriend said you had to be with her it was her birthday or something and I said at the time to you if you didn't come I would I would send some boys over and we would basically kidnap you and take you back over to Brighton in back of a car in the boot so you would so you would appear um and and I suppose i did appear in the end i did appear but only because we threatened you that we would actually kidnap you um and and i guess the point is that conferences and communities and all these events are they a bit of a jolly where everybody just goes along there and it's like a free holiday or are they something that is important still for the individual developer and for customers and so on? So two things.

A, free holiday. This is just to dispel this myth once and for all for everybody listening here, especially everybody listening here that's a client or somebody sitting in his office and thinking, oh, these guys have such a nice life. This isn't free holidays, especially for people who are dependent on their revenue. This isn't definitely free because you actually lose money going there. You're not billable.

You're doing a conference beforehand. You're not billable while you're writing all these slides and coming up with all these things. Because let's face it, writing these presentations and these workshops, this isn't something you do in 20 minutes. Actually, this is a lot of work. So much for the free part. Now, for the other part, it is, I think, absolutely crucial and essential

to the individual developer, once, and to clients and corporations, basically basically in this whole technology area per se. Because, yes, you read a lot of things on the Internet. Yes, you can have 5,000 blog posts per week coming in through your feed that you can peruse and look at YouTube videos on whatnot, but it's actually different to going to a conference, listening to someone like you yourself or me or all the other people who are at UKOG, for example. Listen to them talk. Have discussions with them.

Have discussions with other clients and actually find out, yeah, this is all nice and well, what's being said. But because you're never in the posts online, you never see, well, in reality, this worked like this for us, or we're using it for that. You see marketing slides, you see sales statements, you see wonderful videos about, oh, it's only three clicks to do it, and it's so simple, and everybody can do it, and everything is automated.

Well, yeah, obviously they will not show you the pitfalls. They will not show you the problems. They will not show you the things that are limited, et cetera, et cetera. So these conferences, just in terms of exchanging on all of these things, in themselves are already worth it and the other thing is also I think it really opens the mind to new things you see different you speak to different people you get out of your your your world of this is the way you've been doing it this is the way we

are doing it now this is the way we will doing it now. This is the way we will be doing it. And think about, well, we could do it differently. Listen, this technology is in use at another client as well. They're using it differently. They're using it with something else. They're using it for something else. Why not try it ourselves? Or we have this problem.

How did you solve it? We're facing the issue of the data protection regulations next year. What does the general public have to say about this? How do other people see it? And I think last week, well, not last week, three weeks ago at the OEG, there were several talks about non-techie topics like GDPR, more general things, more management things that affect you.

And I really think people not participating in the community, they will stay in their own small little world. They will do the things the way they do it, the way they've always done it. They will not rethink what they're doing and how they do it. They will not question themselves or their work.

And this is honestly, especially for someone in our business. I mean, look at how many technologies you touched in the last 10 years. I mean, it's an incredible amount. We have to question ourselves all the times. I mean, you said it several times i mainly work with our clinics products yes but i touch dozens of products at the site to keep track of what's happening what could i do and especially to have this this openness of mind hey could i do this differently if yes how using tool? And what could I tell to the client?

Can I actually tell him in good conscience and actually with a base foundation of knowledge, you should do this differently or you should not do this differently because ABC and tell them really, these are the reasons why you should or shouldn't. Because if you stay in one technology, if you always just see one side of the coin,

you will always tell the same thing to your client, you will always solve the things the same way. Well, and in the long run, this will definitely be detrimental to you as a contractor, or if you're at a client, to you as a client, because at some stage, the client will move on and you will be moved into a more like a dead end street. Because you are one of the developers who always did that and who never was open to doing something else. And I've seen this happening at clients with whole teams that were just, I can't put it differently,

boneheadedly following their own way. And all of a sudden, a new CIO arrived with a lot more openness of mind. They changed the corporate structure. They changed the strategy in terms of data analytics. And boom, you're in a department that's basically slated to die. So one of the things, I mean, I used to be involved in a couple of user groups in terms of being on the board and and being being part of the organization and one of the things that struck me was a huge amount of effort of user groups seems to be to justify the fact that there's a user group and to justify the conference end of the year and and you know because people in

the old days people would join up to a user group to get access to beta software or going back to your thing about siebel you, knowledge was not easily shared. And actually, it was a thing that had value and people would join up because of that. I mean, do you think in a way that the model of a conference where you pay money, sorry, a user group where you pay an annual subscription and you go to a conference at the end of the year and you take a week off for that do you think that is redundant now now we've got things like kind of online forums and virtual communities and things like irc and so on do you think that's do you think they're redundant now or or is a different sort of thing really uh we we still need them because

as you said they're still they're still the old guard around and they're used to this and maybe not very open to other approaches. But in the same breath, I have to say, user groups, the classical user groups, need to reinvent themselves. They need to change their approach and not just have the one big conference or some regional forums, etc., etc. I think we need definitely a rethinking of this as well

in terms of the user groups, more proactive, more immediate exchange, more small scale, because not everybody can go to something like Kscope or OpenWorld or UK or UG because as I said you more or less have to take a week off or at least a couple of days and these user groups they will definitely still be around I've seen several of them shrinking in the last couple of years in terms of audience. Others have been growing like crazy. But something I've seen recently in September,

which I found a really, really, really good idea, was something that happened in Poland for a Polish user group in Krakow. They actually had, in conjunction with the German user group, they had a whole busload of German young professionals either just finishing their studies or had just finished their studies, so being at the beginning of their work life, ferried over to the user group, to the conference, and be there for two days as a next generation event. It's punishment. It's punishment.

I mean, you weren't there. Believe me, it wasn't punishment. And so there's a whole busload, I think 30 or 40 or 40 young people who saw this whole thing. And as you said, the young people, they don't come out of the university and say, I want to be a DBA or I want to be a database developer. But they will come normally, 90%, they will come into corporations where this stuff exists where this is the reality of things where they have to be exposed to it you can't you can't just wash away the old and say well we started in uh from a new slate well it would be nice but daily business has to go on. So this was a really, really good approach. And I thought this should be more pervasive.

This should be more common to do. And the other thing is, I think for most user groups, the online presence is woefully lacking. And for many products out there, I mean, even for much of the new stuff, yes, there is stack overflow or whatnot, but a bit more focused. And you mentioned IRC before and the Oracle Developer Community Forum. So Oracle Developer Community Forums, obviously, Oracle curated.

It's there. It's there, stuff and so on. It's really good. But, you know, it's a forum. So it's there. You wait for an answer. IRC, some people might still remember this technology, Internet Relay Chat.

So I think this, again, there's a couple of old-school technologies. I mean, you were using, what's that thing, your platform that is like Telegram and with channels as well. Oh, Slack and things like that. Oh, yeah, Slack. Yeah, exactly. I mean, those tools, they, I think, should be used more by the different communities, not necessarily just user groups, you know, as in organizations, but generally communities for support,

for problem tracking, for whatnot. We're running an Oracle Analytics channel since years, and I know that a lot of projects have been saved just because we exist. And, oh yeah,

we've been running this with Oracle support or on the forums and, and talking to our integrator for weeks and weeks and weeks. And the response was like, yeah, yeah. It's been interesting. Here's the answer. Go and do it.

Uh, five minutes later, you're done. Yeah. Yeah. Interesting. I mean, you know,

I've, I found that, that, that IRC chat chat channel the other week and uh to answer your question yes i had downloaded all the data and then i've downloaded all the history there and and and it's interesting to see analyzing that and looking at kind of i suppose keywords and looking at kind of trends and so on i'll share with you at some point um you know it's interesting to sort of see i suppose the mistakes that are made and actually kind of

like where people are asking the same question over and over again. And, you know, I mean, last one on that topic, actually, you know, so why don't you write any blogs anymore? Honestly, and not to insult anybody listening, but I was getting too many dumb questions. So every time I would post something technically, I would get dozens of emails and comments. I actually turned off commenting on the blog at some stage. Come on. So, yeah, this is more or less the reason why I stopped, because I rather answer very specific questions either in my chat room or on the Oracle developer communities, than just posting stuff out there and then getting questions which basically can be summed up with,

I didn't read what you wrote. I didn't think about anything that was written there. Tell me the five steps to get to my target, period. I mean, I suppose on i suppose on the on the internet there are it is unlimited supply of people who have not read any of the manuals or documentation um and and and that's a hard one i mean i used to be quite a big um i i used to write lots of blogs and and and more doing the podcast now to do something different but certainly answering questions on forums you know you do start to you do start to kind of run out of energy sometimes don't you but but it's an essential thing and i, you know, you do start to, you do start to kind of run out of energy sometimes, don't you? But, but it's an essential thing. And I think, you know, when you

go and look at a new technology, you need that help there really. And I think, you know, virtual communities are sort of good. The conferences are still good. And I had the opinion a while ago that certainly for the big user groups that had lots of these communities of interest and so on, that were quite kind of in a way ossified and and and sort of like maybe representing areas that were no longer so relevant my line was you know almost like stop stop them all and just kind of like say to anybody in the user group if they want to start a meetup or a group if they get more than 10 people they get some funding and let these things kind of in a way really organically form themselves but i am very conscious it's easy to break things it's very

easy to break things and destroy things but things that necessarily come back and and you know analogy i suppose analogy would be again back to the roman empire it's very easy to break up the roman empire but you know i'd say that running water and uh and reading didn't come back to the world until about 500 years later and it was like Blackadder until then so be careful sometimes what you want to creatively destroy and I think that's an interesting sort of area really Yeah exactly I mean to stay with that

I just thought of something because then you would have some people in in robes sitting around a table and basically shouting at each other what did the Romans ever do for us? Yes. Yeah, I mean, these structures were there for a reason.

Yes, but they also sometimes are holding you back. I mean, that's the thing that is maybe only... I'm not saying the one or the other is 100% what you should do. It's just I know from my own experience, how hard it is to keep a community alive. And especially, you know, meetups and stuff.

And then, you know, half of the people don't show up and then you say, well, it wasn't worth it. Should we do it again? It's a, it's actually a hard thing because there's always this inertia in people to

overcome. You have to propose something. So, so they come and, uh, well. Yeah, well yeah exactly so so we can take a break now and we're going to do is we're going to um we're going to go into the christmas part of the special episode now so all right okay so i'll see you in a second um christian and we'll go in and do that so welcome back to uh drill to detail and the special christmas edition

and what we're gonna do now is take a bit of a break from uh kind of heavyweight discussions about kind of uh consulting and and bi and so on and we're gonna actually i'm gonna spring some things on Christian and get his opinion, first of all, on some five things that were topical in 2017. He's not seen these before. And I'm just going to kind of fire them at him. I want his instant reaction.

What is Christian's shoot from the hip reaction to these five statements I'm going to put in front of him or five things. So first of all, Oracle Analytics Cloud. Oracle Analytics Cloud. Oracle Analytics Cloud. Okay, so great product, the best thing they ever produced, obviously. A bit more snarky comment, same stuff that existed as before product for the future I think it's still in its infancy so a lot of people are struggling with what shall we do with this is it worth going there but it has it has a lot of potential simply because of its roots you know where it came from but those are also the things

holding it back but that's a more technical point it's a strength it's a strength and it's a weakness in the same time it's very much based on obie and it's very much a product and i think you can you can read into that whatever you want you know the two ways really i think it's it's you know it's it's a very powerful platform and you've got in there in the metadata layer you've got all the kind of clever stuff there but it's incredible to my mind certainly in this release i was playing around with it's still surprisingly kind of labor intensive really for for doing yes basic administration and and there's a lot of there's a lot of prior knowledge of Oracle technology and Oracle stack that I think it seems you've got to the point where

I couldn't work out how to log in actually first of all and then I realised it was the old WebLogic username and password admin and welcome one and it wasn't the one that I signed up with and it's a lot of stuff in there that you think to yourself how could anybody, it's like reading a book where this new book massively draws on the...

It's like going to see the latest Star Wars film. It might be a good film in its new release, but it makes much more sense and it fits in much more if you've seen the other seven movies in the series, maybe. Wonderful analogy. Couldn't have said it better. That's a nice analogy. OK, so next one, number two. Oracle Analytics, Cloud and S-Base.

Well, for that, I have to say we're kind of lucky because that version actually existed on-premise before. So this is the actual newer Oracle S-Base code base, and this was integrated in Oracle BI already. So again, if you buy the whole shebang, the big one with everything in, then you more or less have the abilities that you have now on-premise

with the S-Base integration, using it as an accelerator, spinning of cubes, et cetera, et cetera. Then obviously you have the S-B s base only version which is more or less s base cubes standalone with data visualization on top of it and again choose your poison what do you want to do with it what do you want to use it for if you just have s base cubes and you have dv on top of it yeah it can sense. I don't see that much of a use case for it. I more see the use case of either the full one, you know, integrated with

relational data with other data sources, like Hadoop sources, like SAP sources, like whatever. And then the other more, let's say, application products that are out there, like PBCS, so the Planning and Budgeting Cloud Service, which, again, is what you have on-premise, but as in applications, but in the cloud. Pure multidimensional storage technology, yeah. I mean, you can move it to the cloud if you want, but I don't see that much added value over what you have on-premise right now.

Okay, okay. Number three, Brexit. Manuel, you're stupid. It's crazy, isn't it? We've gone from being the most, I suppose, pragmatic and sensible nation in Europe to the most unpredictable and the weakest government and the most stupid at the moment. It's interesting, isn't it?

It's amazing. I mean, after Brexit, there were some data dumps available to analyze the voter data. And I think, you know, crunching the numbers there, looking at that the regions that received the most subsidies from the European Union were the ones that most heavily voted for leave, it just blows your mind. And if you push it to the extreme and they're really, really, really sarcastic, what I am, you could say it just shows why direct democracy does not work. No, I know, I know.

Because certain topics just cannot be left in the hand of the average voter. They are way too complex. It's not about, oh, people are being too stupid and Christian is basically advocating dictatorship. No, it isn't. Those are topics that are so complex and with so far-reaching consequences.

You can't ask people to decide on this. They can't spend months thinking about and weighing alternatives and especially when you factor in the whole propaganda bit. And yeah, this is honestly, I mean, without being sarcastic or anything, disastrous, really disastrous. I know. And, you know, again, disastrous, really disastrous. I know. And, you know, again, back to the model of how we are in business, it was predicated on being able to work in any part of Europe. You know, if that's going to change, I mean, certainly I'd imagine that from the date we finally sail off into the middle of the Atlantic, you know, you will not be able to work in the UK doing consulting, I'd imagine.

And equally, any consulting business in the UK is going to have to set up shop in europe or or kind of like only be restricted to its own market yeah it will it will so heavily depend on how these these wonderful negotiations turn out but it will end up being a loss loss situation it will not be win-win it really will be lost loss i know i know okay number four number four star trek the last jedi um have you seen it i haven't i haven't seen it yet are you are you interested i mean you just start trek i mean it's something that uh you know not star trek star wars right star wars the last jedi have you seen that i haven't seen the last jedi yet uh it's definitely on my list because i'm a diehard imperial so yeah several several of my friends have seen it and and uh they said

it's really funny the imperials have the coolest have the coolest tools and toys but they always manage to miss their target which kind of neatly plugs back into our discussion from before you can have the coolest tools but you have dumb ass managing it it just won't work yes i mean i think two points for me on on on star trek the last jedi or even star wars the next jedi one is i can't believe they've got eight films out of this and i was reading the block the plot synopsis on wikipedia Wikipedia yesterday I just thought really this is this is getting really I mean basically everybody who had a certain position even I think as a spoiler you know I think one of the main characters if you're going to see it but one of the main characters turns out

to be a bit nasty you know and it's like really I mean this is the poster we all know. Yes, exactly. And the second one is, I've forgotten the point now actually, but it just strikes me as, I don't know, it's not for me really. It's eight films in now. It has gone from something that has developed an almost religious character into something that's just being built for commercial purposes, which kind of is basically like a religion started off as a good idea believed in it now it's just something commercial

well how comes the the rebels are the good guys because actually you know what you've got here is a well-ordered society um where where basically there's a bunch of people who start blowing things up all the time and and kind of ruining everything and and you know surely surely america going to the cinema to watch uh to watch the last jedi can you not see the irony in in the kind of people you're celebrating are effectively terrorists and it strikes me as i don't know it's an interesting interesting kind of like lack of maybe self-awareness yeah but if you talk about the americans they're basically celebrating their own their own history you know uprising against the empire so for them you have to say it doesn't make a lot of sense it went wrong for that point almost didn't it really so uh anyway um the last one is the last one the list of five things here is oracle autonomous database cloud service

um and the uh and i suppose no more dbas i mean what's your view on that? The marketing, I suppose the marketing and messaging around the autonomous database service that came out recently. Yeah, right. You can sum it up with a very sarcastic, yeah, right. It will definitely go that way. And as you said, the DBA will more become of an administrator, making sure that thousands of instances run and whatnot, instead of going and twiddling with features of the one or the other. But that product in itself, it's just the first step. It's a baby step.

And it's good. It's a proof of concept, basically. For me, it's nothing but a proof of concept that they put into a product. Yes, it works. Yes, it can be used. I don't know whether anybody will buy this, except for a couple of clients trying it out anyways. But, yeah, it's a first step. I think it's a triumph of marketing really in that that oracle were quite late i suppose to the database as a service and data warehouse as a service market and and you

know there are companies like snowflake already i didn't want to mention them by name but no no and so there are companies out there been doing this but oracle i mean oracle and this is not this is not a snarky comment they are very very good at marketing. And to take something that is a complex platform and then use machine learning in the future to make what is now arguably unnecessary complication automatic, and then to sell it as being the first of the autonomous databases, I mean, hats off to them, really, for taking what is a position where they're several years behind the competition and making it out that actually this is kind of groundbreaking. And, you know, that's why Larry's got a yacht and you and I are kind of like, you know, still working really.

And marketing, I think marketing is seen, you know, sometimes I call it the colouring in department, you know, when I'm being a bit sort of disparaging, but marketing is so important. And it's actually what it's all about is it's about finding a problem. It's about finding a customer with a problem, identifying the problem, solving the problem for the customer and doing it in such a way that they are motivated to actually spend some money or do something. And you know what? That's what those companies are good at doing, really. You know, it's techs and specs and all that stuff is one thing. But you've got to solve a problem for somebody.

And that's what analytics is about. And it's what all those companies have done. And it's why they're successful. Yeah. So, Christian, you're a big advocate of of best practices that you find on the internet and as as as robin as robin and i have have many times kind of ribbed you but let's imagine now that you are starting work as the director of best practices at oracle and you have someone sent you three best practices that you that they found on the internet your job is to either say they're good or they are as i put in in the notes here low roll bollocks which is an

english phrase which means uh yeah anyway so the point is these are three best practices here are these are these actually best practices or are they or are they kind of misconstrued and your job is to kind of to tell us which one is correct and to give us maybe your view on it so the first one is a document i found on the internet which was an oracle oracle best practices document from a few years ago which was top tuning recommendations for obie and it had in there it had a series of seven recommendations these are the seven things you should do in order to tune your obie system and the first one was tune the operating system. Second one was tune Oracle WebLogic server parameters. Then it was tune the Java virtual machine, tune HTTP servers, and tune the database.

Is that correct or is that mistaken? Each point in itself, if you have an issue in that specific area, is valid. As a total, if you just say you have performance problems, take this document, go through step 1 through 10, and you will have solved your performance issues, this is totally wrong.

This is a 1,000% misconception, but again, comes down to people just taking things literally and as they read it, not thinking, I'm just going to go and tweak something on the HTTP server. I'm going to fiddle around with my Apache or I'm going to just up some memory settings in the Java virtual machine. Most likely, you will fiddle with something that you do not need to fiddle with. You may end up actually making things worse for you because you don't know what you're doing,

why you're doing. And in the end, yeah, you'll probably do this, put a lot of effort into it, put a lot of time into it, and then manage will come and say, well, we didn't actually get any benefit out of it. What the hell? Why did you do this?

This is a load of rubbish that you did this document is rubbish let's switch to a different technology so best practices and especially when it comes to performance and this is not just oracle this is with any technology when it comes to performance this is the point where you have to think where you have to analyze what happens where when why where is the time spent where's the cpu spent where is the ram spent and just going through the document and ticking off boxes is pointless yeah exactly i mean we we have this in same debate my old company about about best practices and people people like best practices you know they always do the joke that the uh the best way to get a conference presentation accepted anywhere was to put in best practices for whatever product.

Why do you think I'm running a word practice series on so many different topics? Yes, people love best practices and they also love best practices being debunked as well. But I suppose people want guidelines and they want maybe what are good practices with context and and so on i mean but there was a practices yes yes i know exactly and to that point you know when uh robin joined my company a while ago we again jokingly i jokingly said to him he would be our director of best practices to just to wind him up and um but i said to him look actually look there's a reason behind the reason people ask for these things not because they want necessarily to follow something blindly but they want what are our recommendations around certain areas and actually got him to to write

a paper at that point which was what we called the good practices for oracle bi at the time and they're there to try and address that and there was a best practice in there was a good practice in there that uh is the second one i'm going to write i'm going to read out now and get you to comment on this okay so um aliases can be a useful way of managing the name and organization of physical tables blah blah here adopts a consistent naming convention for aliases typically the convention is to and it goes on there and talks about prefixing with dim and facts and whatever so question to you are aliases mandatory and best practice or good practice in uh it can be alias aliases is basically you have a physical object like a database table uh and you instantiate it twice under a different name to use it for example once as a i don't know customer address

dimension once as a corporate address dimension, once as a address dimension tied to a purchase order, et cetera, et cetera. So it's basically reusing existing things, making nice little, as I always say, Lego bricks out of it and plugging them together. And it is good to make things readable. Yes.

It is good if you want to have role-playing dimensions, if you want to have role-playing facts so that you don't have to duplicate things in your database layer, etc., etc. That said, it is not an absolute yes or no. It depends on what you want to do with it if you have data models that are built for analytical access that are perfectly built data marks let's say you're in you're in a situation where you have a department that prepares the data marks and they're really

into dimensional modeling and they have perfectly uh defined and named text dimensions, whatever. Adding aliases on top of that is redundant because it's already done. On the other hand, let's take the other extreme. Let's take the case where we're more doing, let's say, rapid prototyping or really getting stuff out to the client as quickly as possible, and you just have flat tables or something in Hadoop that has no structure, that just has 500 different attributes. Some of them are facts, some of them are dimensions, and you want to make sense of this so people can analyze the data. Well, then obviously you will not just dump out 500 attributes to the front end

and say, well, have fun and do something, we still want to guide them. Not everybody is a data scientist. Not everybody is an analytical consultant and knows how to work with data out of the box. So you need to structure things. So you would just take this flat table with these 500 attributes, alias it several times, and say, whoa, you remember entity relationship diagrams, you know, back in the 80s?

Entities actually have a point. Something is the customer. Something is the customer case or his purchase order or his insurance claim. Something is the account. Something is the product. Something is time. Something is the account. Something is the product. Something is time. Something is the fact data that we're actually counting.

So you split up these entities into alias and then you start modeling on top of it. So it does make sense to use them in these cases. It does not make sense at all if you have a perfect data model underneath that's used for that and only that purpose. Yeah, I think it's a couple of things that come out of that. One is, obviously, there's the sort of slightly jokey thing about just following a rule that says you should always use aliases.

And the same thing about you should always turn off your logging if performance is bad. There are things where someone has written it one time, it might possibly have applied in a certain circumstance. It might not even have been that. But it's following it slavishly but i think the wider point is about things like kind of i suppose um coding standards it's about organizing your thoughts it's about kind of if you're going to build a data model for obie or other kind of bi tool make sure it's clear in your own mind about the role it's playing and the various entities in there and so on you know in a way that's what that is what you know we used i used to always advocate um using aliases because it made people think about the role of the tables and so on but but just copying something is is you know is true

for that getting people to think about it definitely and that's the reason why why i always say to people listen if you have a problem comprehending what that thing does or how it's built, there's nothing better than taking a sheet of paper and a pen and drawing it and putting names on it. And then you can institutionalize it with aliases, for example, and then your business model. But just go back and think about it. Draw a picture. It doesn't matter what you're doing. Draw a picture. Because it's different than clicking around in a tool. A piece of paper and a pen gives you 50 times as much comprehension as saying, oh yeah, I'm going to draw some joints here

and I'm going to write some code there. Okay, okay. And the third one, the third best practice is more like a design guideline. Imagine a certain gentleman who lives in America who argues with people on the Internet. He would say every BI project should be an agile project. What do you think? I don't know who you mean. Well, let's say definitely the old school way of doing things like waterfall style. We do a project for three months and then we give you something

and you give us feedback on it yeah, that just doesn't work, that just doesn't work projects have to be I don't say they have to be agile in the sense of the term and you have the agile methodology but anything that goes

into rapid prototyping, rapid development and basically getting results to the people as fast as possible and then discussing based on these results with them is is the way to go let's put it like this i've seen a drawing a couple of years back yeah we're going to develop a car and the old school let's say waterfall approach is phase one we have the chassis phase two we put the motor in it phase three uh we put some body work to it and phase four we attach wheels and in phase five you can drive it that's waterfall uh a more agile approach is phase one you have a skateboard phase two you have a

controller phase three you have a bike etc etc so everything is usable not like the first version where well you have to wait until the end to make it usable so get things to your people as fast as possible talk with them talk about the data they're them talk about the data they're seeing talk about how they're using it if you can do this in in in a couple of minutes or in a couple of hours instead of taking months and huge requirement documents you will have different discussions with them you will actually have proactive and and productive discussions rather than saying you didn't do what i specified and and uh then the

reply being well you didn't tell me to specify this exactly it just isn't isn't leading you anywhere because then they will just say you know what give me the data dump and i'll do it in excel period and you've lost them i always think you know with agile projects it's interesting you've got it i think you've got one thing i've noticed is that you can say agile and you can be agile. And there is an agile methodology or method that people don't tend to follow. And I always tend to find with agile projects that it's a bit like communism, you know, in that whenever an agile project fails, it's because you weren't using... Well, when you look at any country that kind of failed on Russia or anything at all

where they were a communist country and the economy fell apart, the argument was always they weren't communist enough. you look at any country that kind of failed on the russia or or anything at all where where where they were a communist country and the economy fell apart the argument was always they weren't communist enough they weren't pure communist that sort of thing and obviously agile was a bit like that in that um in that you know when a project fails an agile project the argument is it wasn't agile enough and the problem is with communism you know it was like that and then everyone died at the end and i kind of think it's you you know, with Agile projects again as well, I mean, I suppose the point with it is, is that actually it might work if you all do Agile perfectly

in the same way that a communist society works if you all work, if you're all true communists. But people aren't. People kind of, they say they'll be Agile, but they actually still want to have a fixed set of requirements at the end. Or they want you to, in the same way that communist societies kind of failed

because people were lazy, people were kind of greedy, people were whatever. You know, with an agile project, it's all well and good. But people tend to kind of like fall back on the old ways of doing things. So either you basically completely re-educate society

to be agile, or it's just going to be doing a normal project, but you do it without any testing or on half the budget really. So last thing for you because I haven't been going quite well now, which is great, is predictions for the future.

So predictions for 2018. Okay, so I'm going to ask you three things. So what do you think is kind of, what are people talking about about what happened next year but actually very unlikely to happen? And then I want to get your take is kind of what people were talking about, about what happened next year, but actually very unlikely to happen. And then I want to get your take on what is most likely to happen.

And then maybe a sort of funny one at the end, you know, what is going to happen, but it's a real disappointment really, you know? So first of all, what do you think is everyone's prediction for next year that they say is going to happen, but actually in your experience won't? Oh, that one is definitely uh this whole thing i

mean it's it's a it's been around since a while but it's it's still it's still being flom but uh everyone is being empowered to be a data scientist you had a presentation on that once and and you called it the um the citizen data scientist? Yeah, exactly. I more or less say, well, it's actually the peasant data scientist because this will not happen. Not everybody is a data scientist. Not everybody is cut out to be a data scientist. Not everybody should be a data scientist.

I always say you have, for example, people very clever in statistics. Yeah, okay, that's a statistician. That's not necessarily a data scientist, because a data scientist also has to be aware of how to work with data and not just saying, I'll run my algorithm. So I think this is one of the big misconceptions, and everybody's putting that on their CV, obviously.

I mean, how many CVs do you see floating across your desk that says data scientist, and then you look? Oh, it's data engineer. Data engineer. Data engineer is the new data scientist. Yes. Both. And then you see, well, either he's doing statistics,

or he's a DBA, or he's an ETL guy, or whatever. And you're like, yeah, okay. Well, you're one of those. Okay, okay. DBA or he's an ETL guy or whatever yeah okay you're one of those okay okay so what do you think what do you think is what if you had one prediction for 2018 that you think will happen and you know what is your what is your prediction that will be

in the news or be topical sort of for next year very very definitely GDPR and this will and and in in britain in britain with brexit we've completely missed that but what is gdpr and why is it something to be very um aware of really yes so so just very quickly because people people who are interested in this can can also listen to your your podcast that you had with Carson because he actually went into a lot of detail. So GDPR are the data protection regulations of the European Union, which actually, and this is something people missed, that they are actually in effect already.

It's just that as of May next year, you can be, well, dragged in front of a court and punished for violating these guidelines. And generally, you have to be able to prove where is your data stored, what's being done with your data, who is using it, who did it when. You have to be able to get rid of the data, so the right to be forgotten is built in, for example. And if you use the data in and this is

where the killer actually comes in for analytics is the right for the client that he can tell him okay your data was used analytically for example in some algorithm in some, in some algorithm, in some analyses, in some calculation, what were the rules lying behind it? What was the algorithm? How did you get to this result? If you have a credit scoring of 42, why is it 42 and not 27? You have to be able to prove this. And this actually, if you take all this together, this is actually what people should be concerned about with all their decentralized data silos lying around. Somebody doing R somewhere, another one doing Tableau, then having some Excel data pumps and Excel sheets flying around. You lose control of the data. If you cannot trace and be able to tell the customer, okay, this is where the data export happened,

and then we put it into, I don't know, a source or Excel or somebody ran an R algorithm on top of it. Yeah, but what did they do? What rules did they use? What procedures did they use? And how did you get to this information? You will basically, by default, lose the lawsuit because you've lost the control, you've lost the traceability. You cannot tell him why did you get to this result.

So I think this is, I said we had this move away from centralized governed IT because speed to market wasn't fast enough and more to departmental stuff. Again, you know, back from the web front and down to the desktop. This will actually be a quite tricky point for corporations next year when people start realizing, and I'm pretty sure that there are several lawyers' offices already in the starting blocks with predefined lawsuits and saying oh yeah, if you have something

like it happens in America if you have case ABC just come to us, we got it covered already, we'll take a part of the money that we win and then we'll do it and I'm pretty sure this will, this will happen as well because I guess the lawyers who will be attacking the corporations are a lot more ready than the corporations who should be actually

getting ready. And the other point that, that Carsten made last week was what about companies that use things like neural networks and these very, these kinds of self learning, you know, machine learning kind of things where, where you can't even explain even explain i mean going back to the days of remember trying to explain remember someone tries to get you to document how an rpd worked years ago and and

because it wasn't a simple data model it was quite hard to do that because you had all these different i suppose multi-dimensional parts to it and so on and and you know and that that was it was hard to explain that imagine trying to explain how a neural network's when actually in some respects actually it isn't even understandable because it's been actually organically developed by the machine learning algorithm how you're going to how's that but that's the thing but that's the thing i've stuck more with the analytics world because this is what we've been talking about uh and this is what what most people are are uh are interested in uh right, is the burning topic that's on their hands at the moment.

We haven't even talked about AI, as you said. But this, I mean, obviously, the law hasn't foreseen this. It has to be said as it is. The law hasn't foreseen this, that there is no real flexibility in that regard. A, the law has to change, and quite quickly. They will have to adapt some things there. And B, if you're very, very much at the forefront of this, and someone basically wants to attack you on that you're you're vulnerable i don't know how

they will manage it i mean there's there's five there's five no six months there's six months left now uh so it's it's going to be interesting how this all turns out uh but it will definitely burn some people which just aren't prepared or taking it uh well too lightly yeah okay and the last thing here is your prediction for the thing that we don't want to happen but unfortunately probably will um everybody will still keep talking about buzzwords instead of doing it so uh you you will you will still have a lot of of people saying all the right things, but in fact doing all the wrong things, but just like Agile. Oh, we're doing Agile waterfall.

What? What do you mean Agile waterfall? That's a contradiction of terms. Yeah, I guess there's always something, putting it more human instinct, there's always something that's more interesting from a technology perspective to look at. And I think there's always value in trying new things out and experimenting and so on. But one thing I've learned from the work I've been doing the last year is trying to understand the core purpose of analytics.

What is it there for? What does it do? And effectively, you've got a problem to be solved, and it helps you solve that problem in a more effective or more informed or more a more kind of uh rewarding kind of way and and so it's understanding that all these things they're there to solve a problem they're not there in themselves as a kind of the technology isn't the purpose of it really and that's why sometimes excel can be the right thing it's sometimes in process done to be right things but it's about understanding the problem you're going to be solving and it's about culture as well and and i think that you know that's a kind of nice way i think of of ending the conversation really i think that we've i think the message i've got

from you really in this in this kind of conversation is is that all these things are there but it's about understanding the problem understanding the customer um and doing the right thing by them and trying to understand what they're trying to do and what they're trying to solve and then solving that really. And that's pretty why you've been a survivor for the last 15 years. Yeah.

Somebody has to do it right. Yeah, exactly. Exactly. So Christian, thank you very much for coming on the show. It's been brilliant to speak to you and have a good Christmas

and hopefully you won't have to sit through that awful awful British program about the dinner thing with the Oh, dinner for one. Christmas in Germany, isn't it? That in Britain nobody has any idea why you watch that.

It's awful, but it's a massive institution in Germany, isn't it? It's like the counterpart for everybody in the Anglo-Saxon culture is all about the sound of music. And to us, it's basically an insult. So this is our way of getting back at you. Exactly. Christian, great to speak to you and take care is our way of getting back at you. Exactly. Christian, great to speak to you. And take care and see you next year. Thanks for having me.

And happy Christmas, everybody. Thank you.